Original title: Fraud Detection with Positive and Unlabeled Dataset using Interactive Learning Technique
Authors: HASSAN, Mobin Al
Document type: Master’s theses
Year: 2025
Language: eng
Abstract: This study is the focus on positive and unlabeled (PU) labeled learning with a new technology being the interactive learning, which solves the problem of training machine learning models with messy, unlabeled real-world data, a common example is financial transactions where labeling is expensive or impossible. PU learning, on the other hand, is quite unlike conventional well-balanced data sets because it involves a small number of positively labeled examples (or "positives") and a huge number of instances which are unlabeled. For working with tabular data, the research suggests integrating a tree-based algorithm into the existing reinforcement framework instead of commonly used neural networks in existing algorithms. The research aims to transform fraud detection systems into more accurate and efficient ones by implementing interactive learning, which will consequently help overcome the rabid demand for data quality and labeling becoming a barrier in real situations. The empirical results will give an insight as to how these methods are applied and what could be the potential of fraud detection system improvement.
Keywords: Artificial Intelligence; Data science; PU learning; semi-supervised learning; weakly supervised learning
Citation: HASSAN, Mobin Al. Fraud Detection with Positive and Unlabeled Dataset using Interactive Learning Technique. České Budějovice, 2025. diplomová práce (Mgr.). JIHOČESKÁ UNIVERZITA V ČESKÝCH BUDĚJOVICÍCH. Přírodovědecká fakulta

Institution: University of South Bohemia in České Budějovice (web)
Document availability information: Fulltext is available in the Digital Repository of University of South Bohemia.
Original record: http://www.jcu.cz/vskp/77560

Permalink: http://www.nusl.cz/ntk/nusl-695406


The record appears in these collections:
Universities and colleges > Public universities > University of South Bohemia in České Budějovice
Academic theses (ETDs) > Master’s theses